Explore My Creative Side
Discover my artwork - where technology meets creativity and stories come to life through colors and shapes.
Visit My Art Gallery
Harleen Kaur
I'm a
Master's student in Computer Science at Virginia Tech, specializing in Machine Learning and Data Analytics. With a knack for transforming data into insights and ideas into impact, I blend technology, creativity, and leadership to shape meaningful solutions.
Actively Looking for Job Opportunities in Full Stack, Data Science and Data Analyst Roles
Academic Projects
ML-Based Rental Price Prediction System
Overview: Developed machine learning models to predict rental prices and classify properties into market segments, helping renters and property owners make data-driven decisions.
Key Features:
- Implemented supervised learning (Linear Regression, XGBoost, LightGBM) and unsupervised algorithms (K-Means) achieving ~90% accuracy
- Enhanced prediction through dimensionality reduction (PCA), feature engineering, and geospatial data enrichment via Geopy API
- Validated model performance using cross-validation, hypothesis testing (T-tests, F-tests), and statistical validation techniques
- Discovered market patterns through clustering analysis and association rule mining
Python
scikit-learn
TensorFlow
Pandas
NumPy
Matplotlib
Seaborn
Emotion Classification using NLP
Overview: Created a text-based emotion classification system using the Hugging Face "Emotion Dataset" to categorize text into six emotions: Sadness, Joy, Love, Anger, Fear, and Surprise.
Key Features:
- Performed extensive data preprocessing including lowercasing, tokenization, stopword removal, lemmatization, while preserving emotional cues
- Implemented multiple feature extraction techniques: Bag of Words, TF-IDF (with feature selection using Chi-square), and pre-trained embeddings (Word2Vec, GloVe, Sentence-BERT)
- Trained and compared multiple models including Random Forest, SVM, and XGBoost
- Achieved best performance with Word2Vec + XGBoost (68.90% accuracy) versus Sentence-BERT (66.95%) and traditional approaches (59%)
Python
NLP
Word2Vec
BERT
TF-IDF
XGBoost
SVM
Random Forest
Implementing Naive Bayes & Decision Tree on FMNIST Dataset
Overview: Implemented a Naive Bayes classifier from scratch and a Decision Tree model using Scikit-learn to classify clothing items from the Fashion MNIST dataset, focusing on distinguishing between Trousers and Pullovers.
Key Features:
- Built Naive Bayes classifier from scratch, computing conditional probabilities and posterior probabilities without using ML libraries
- Developed a Decision Tree classifier (max depth = 10, Gini index) for performance comparison
- Manually implemented evaluation metrics including accuracy, precision, recall, and ROC curve without using existing libraries
- Experimented with various binarization thresholds (100 to 200) and analyzed impact on classification performance
Python
Naive Bayes
Decision Trees
scikit-learn
FMNIST
ROC Curve
Feature Preprocessing
Interactive Data Visualization Platform
Overview: Created a web-based analysis platform with dynamic visualizations enabling users to perform complex data exploration without programming knowledge.
Key Features:
- Built interactive dashboard with dynamic filtering, feature selection, and statistical analysis tools
- Implemented comprehensive visualization suite for distribution analysis, relationship visualization, and categorical data exploration
- Integrated analytical capabilities for outlier detection, PCA-based dimensionality reduction, and normality testing
- Deployed containerized application to Google Cloud Platform for worldwide accessibility
Python
Dash
Plotly
Matplotlib
Seaborn
Docker
GCP
E-Commerce Platform: Java & Angular Web Application
Overview: Engineered a robust, scalable e-commerce platform that revolutionized online purchasing experiences.
Key Features:
- Secure payment integration with Stripe, ensuring seamless and protected financial transactions
- Advanced backend optimization with Java, reducing database query response times by 40%
- Comprehensive Angular frontend delivering responsive, intuitive user interfaces
- Automated customer communication workflows enhancing user engagement and retention
- Implemented complex SQL query optimizations for efficient data retrieval and management
Impact: Developed a high-performance e-commerce solution that significantly improved transaction speed, security, and overall user experience.
Java
Spring Boot
Angular
SQL
Stripe API
RESTful Services
Work Experience
Full Stack Engineer
Fidelity Investments
August 2021 - August 2024Data Engineering & Analytics Projects
Cloud-Based ETL Pipeline
- Designed and implemented a comprehensive data transformation architecture using Spring Batch, AWS S3, and Snowflake, creating a scalable system that replaced legacy flat file processing
- Implemented statistical validation procedures and data normalization techniques to ensure data quality and consistency
- Achieved $2.5M annual cost savings through architecture optimization and process automation
Spring Batch
AWS S3
Snowflake
Java
Statistical Analysis
Real-Time Data Processing System
- Led development of an end-to-end data pipeline that processed 50K+ records per minute from mainframe systems to cloud environments
- Engineered data transformation workflows using Python, SQL, Spring Boot, and CDC replication with Kafka integration
- Implemented statistical analysis methods to validate data integrity during high-volume migrations
Python
Spring Boot
SQL
Kafka
CDC
AWS
Oracle
Automated Data Validation Framework
- Created a robust validation framework achieving 90% test coverage for critical customer account workflows
- Developed rule-based validation checks and statistical thresholds to automate quality assurance
- Designed modular components that could be reused across multiple data pipelines
Java
JUnit
Cucumber Framework
SQL
Statistical Methods
System Optimization & Infrastructure
Production Performance Analysis
- Conducted comprehensive analysis of production systems by extracting patterns from error logs and performance metrics
- Identified 50+ critical system bottlenecks through data-driven investigation
- Implemented SQL query and stored procedure optimizations that reduced latency by 40%
SQL
Performance Tuning
Log Analysis
Statistical Methods
Datadog
Perf Testing
Cloud Infrastructure Automation
- Architected containerized data pipelines using Terraform, Docker, and Kubernetes (EKS)
- Implemented CI/CD workflows that increased deployment frequency and reduced manual intervention
- Improved data processing efficiency by 60% through automated workflow optimization
Terraform
Docker
Kubernetes
AWS EKS
CI/CD
Database Architecture Analysis
- Conducted data-driven proof of concept comparing CockroachDB and Oracle performance metrics
- Analyzed read/write speeds and transaction processing capabilities under various workloads
- Provided evidence-based recommendations for database architecture to support high-volume financial data processing
CockroachDB
Oracle
Performance Testing
SQL
API Development
Mutual Funds Domain
- Developed a dynamic Rule Engine API with multiple parameters for accurate fund selection and streamlined rules creation
- Maintained comprehensive Swagger documentation for RESTful APIs to ensure seamless integration
- Designed and implemented microservices architecture serving 10K+ daily users with robust Java backend services
Java
Spring Boot
REST API
Swagger
Microservices
Datadog
AWS
Docker
Frontend Integration
- Debugged and enhanced frontend applications for mutual fund management using Angular and React
- Implemented feature enhancements that improved user experience and application performance
- Conducted UI testing using Selenium framework to ensure cross-browser compatibility and responsive design
Angular
React
TypeScript
JavaScript
Selenium
Summer Analyst
Fidelity Investments
May 2020 - July 2020Feedback Analytics System
- Engineered a data collection and analytics system for 60,000+ employees using SharePoint
- Implemented A/B testing methodologies to optimize feature placement and user engagement
- Developed interactive dashboards using Power BI and Tableau to transform feedback data into actionable insights
React
TypeScript
Power BI
Tableau
A/B Testing
SharePoint
SharePoint Performance Optimization
- Reduced SharePoint page load time by 40% for all employees by standardizing JavaScript and CSS file references across all sites
- Improved user productivity and increased platform engagement through optimized web performance
SharePoint
JavaScript
CSS
Web Performance
Skills & Technologies
Programming Languages
Frameworks & Tools
Cloud & DevOps
Security & Compliance
Database & Big Data
Data Science & ML
Statistical Methods
Data Visualization
Certifications
AWS Certified Cloud Practitioner
Database Certified SQL Associate
Information Security (OWASP)
Reach Out
About Me
A storyteller at heart and an engineer by trade, I thrive at the crossroads of creativity and technology. Whether it's designing intelligent systems, leading impactful initiatives, or bringing ideas to life through code and art, I believe in innovation with purpose.
Currently, I am a Master's student at Virginia Tech, specializing in Machine Learning and Data Analytics, with three years of experience at Fidelity Investments as a Full-Stack Developer. I've built large-scale data pipelines, optimized cloud infrastructures, and engineered solutions that saved millions—earning me two Impact Awards. My leadership journey includes heading Learning & Development at Fidelity, mentoring teams, and driving organization-wide upskilling initiatives. Now, as a Graduate School Assistant at Virginia Tech, I blend research, strategy, and innovation to support academic growth.
Beyond tech, I paint, teach, and design. As the Social Media Head of E-Cell, I nurtured my love for branding and storytelling, crafting digital experiences that inspire. I also enjoy mentoring students, breaking down complex topics into engaging lessons.
When I'm not coding or strategizing, you'll find me bringing stories to canvas.
Email:
Phone:
Location:
Arlington, Virginia, USA
Degree:
MEng in Computer Science